Introduction
The key to public health, said Theodore D. Woolsey, director of the U.S. National Center for Health Statistics (NCHS), was data systems. It was June 1968, and Woolsey was welcoming more than 400 vital statistics experts to Washington, D.C., for the twelfth national meeting of the Public Health Conference on Records and Statistics. In his opening keynote, Woolsey explained that one of the conference’s main goals was to discuss the various metrics officials needed to improve public health. But just as important, he added, would be discussion of the country’s network of data systems – the infrastructure that turned anecdotes into data, then data into policy recommendations. As Woolsey explained, public health statisticians in the United States would soon find themselves in a dilemma. On the one hand, a recent wave of federal health legislation was about to increase demands for hard evidence; on the other, the new era of stringent government budgets would soon make gathering such data more difficult.Footnote 1 Reconciling this tension required the ‘utmost in ingenuity’, Woolsey said. And though there was no question that computers would be useful in this realm, he added, he cautioned his colleagues to tread carefully around technological hype. ‘We sometimes tend, mistakenly, I think, to put the emphasis on acquiring the computer capability’, Woolsey said, ‘but no computer can create data where none exists’.Footnote 2
Woolsey’s warning was prescient. Facing both demands for more data and a deluge of budget cuts, public health officials beginning in the late 1960s set out to transform the way the U.S. government collected death information.Footnote 3 But the process did not go smoothly. In their efforts to revamp the country’s vital statistics infrastructure, officials spent decades trudging a ‘complicated’ path.Footnote 4 At the heart of the challenge was the issue that Woolsey foretold: computerisation was inevitable, but computers could only do so much, and what they could not do was perform the critical first step of gathering death information, which required communication between local health officials, physicians, coroners, medical examiners, and funeral directors. Even by the early 2000s, when U.S. states finally began implementing electronic death registration systems (EDRS), issues obstructed death data collection.Footnote 5 And though modernisation attempts since have been largely successful, with strides made to improve and speed up death data collection, impediments continue today. Well into the twenty-first century, EDRS tools continue to introduce new issues, leave legacy problems unresolved, and frustrate many death data workers.Footnote 6
There is no shortage of scholarly literature on mortality statistics, but the historiography on U.S. civil registration systems is minimal.Footnote 7 Two recent books address the topic best: one on births, one on deaths. Historian Susan J. Pearson’s The Birth Certificate describes the efforts of reformers and bureaucrats who believed that population-level birth statistics would lead to better health outcomes. Around the turn of the twentieth century, this motley crew agitated for better birth registration systems and worked to promote their value.Footnote 8 Similarly, in Count the Dead, historian Stephen Berry traces the formation of U.S. death registration systems. Berry argues that the widespread adoption of death registration areas in the first half of the twentieth century was monumental, with the act of counting deaths leading to the global doubling of life expectancy.Footnote 9 But Berry’s book is teasingly concise. It focuses predominantly on the arduous origins of the country’s death registration systems, omitting the challenges the infrastructure faced just decades later.
Expanding the chronological scope of U.S. death registration history is worthwhile, because doing so offers insights into how the country’s public health infrastructure responded to a rapidly changing information technology landscape. Here, there is some historiographical precedent. Most relevant is work from Edward Higgs, which explains that England’s General Register Office (GRO) became a site of technological innovation, but this was not a case of technological determinism, Higgs argues. New machines and techniques are introduced only ‘because they fulfil a need, or solve a problem’, he writes, and it was never inevitable that GRO officials would recognise shaky or incomplete vital statistics as such.Footnote 10 In this article, I tell a similar story, exploring a field that identified a handful of interlocking problems and treated new technology as the solution. In the late 1960s, U.S. public health officials began their efforts to reimagine and ‘reengineer’ the sluggish system they used to collect information about deaths. Far from being a tidy tale about a paper-based process that needed simple upgrading, this history shows just how challenging the modernisation has been. Building on the work of infrastructure scholars like Geoffrey C. Bowker and Susan Leigh Star, I take seriously the political landscape of vital statistics modernisation.Footnote 11 Despite scant resources, conflicting demands, and an unreliable cast of software users, top U.S. public health statisticians slipped into what historians Lee Vinsel and Andrew L. Russell have called an ‘innovation delusion’, prioritising the promises of new technology and, by doing so, intentionally or not, devaluing the work of maintenance.Footnote 12
Throughout the article, I pay particular attention to the topic of data entry labour, the work needed to make digitised death information actually feasible. Though many public health officials involved in mortality statistics computerisation spent decades pleading for data entry labour to be treated as a serious concern – with some even anticipating that reengineering the ‘long chain of individuals’Footnote 13 would pose ‘a major disruption’Footnote 14 to the data – I argue that the hegemonic positioning of EDRS as both a beacon of technological progress and an antidote to dwindling budgets largely obscured these warnings. As a result, the mere premise of EDRS was championed for its supposed ability to improve the integrity of death data while also ameliorating data entry burdens. But these issues, I argue, were in tension, leading to frustration, shoddy data, and significant rollout delays. Taken together, the article emphasises the importance of human labour in the state’s collection of death data and interrogates the premise that vital statistics infrastructure can be improved through investments in technology alone.
Evaluating the problem
The modern landscape of computerised U.S. mortality statistics originated in the 1960s with the creation of the National Center for Health Statistics (NCHS), which merged the National Office of Vital Statistics with the National Health Survey.Footnote 15 At the time, the demand for better, faster mortality statistics was growing rapidly, as the country processed the wars and epidemics of the previous decades, and as the widespread introduction of cars on U.S. roads proved deadly. Simultaneously, vital statistics officials recognised the need to ‘keep abreast of rapidly evolving automation capabilities’, suspecting that modern computing infrastructures could create new incentives to gather and share data, in addition to bolstering the death data already collected.Footnote 16 This latter need was critical. According to published notes from a 1966 meeting of the American Association for Vital Records and Public Health Statistics (AAVRPHS), an organisation that represented state and local vital records offices, the association had grown ‘deeply concerned’ with the validity of records and with the quality of the statistical output.Footnote 17 The group’s president, Leland E. Aese, warned that these concerns would only expand as the country’s health programs required more, and more sophisticated, information. Additionally, the introduction of new and increasingly capable computers, Aese wrote, ‘has generated a chain of circumstances that require that we take a harder and longer look at what we are doing and why’.Footnote 18
Why was the data so concerning? In the eyes of Forrest Linder, then director of the NCHS, the problems were threefold. First, and most importantly, state offices that managed vital records were not getting ‘a proper share of the grant money … made available for health programs’, Linder claimed.Footnote 19 The other two issues stemmed from these budget woes: staffing and training. The field struggled to find personnel, Linder said, and it could do far more to develop retainment and training programs. These were fundamentally human issues, not technological ones, but Aese nonetheless asked attendees to take seriously the premise that computers might play a role in addressing these concerns. The group responded eagerly to Aese’s charge. By the end of the meeting, AAVRPHS members even decided to task their archives committee with redesigning the cover art on the association’s self-produced history volume. ‘It was the opinion of the group’, the meeting notes said, that the design should include ‘a quill pen and a computer’.Footnote 20
The AAVRPHS members were not alone in their excitement about computers. By the end of the 1960s, discourse about automation buzzed to the highest rungs of government and industry, and in the realm of vital statistics specifically, the NCHS’s computerisation initiatives were already well underway. The NCHS focused its early death data computerisation efforts on non-registration projects, however, steering clear of the labour-intensive process that involved a large cast of characters. Death registration relies heavily on the work of a funeral director, who needs to obtain a decedent’s personal information from loved ones, coordinate with the certifying physician or medical examiner to collect cause-of-death information, then file the paperwork with a local registrar within a few short days after the death.Footnote 21 By contrast, nosology would be far easier to automate, NCHS officials suspected, because it was done by only a small number of highly trained experts. Vital statistics computerisation thus began with efforts to automate the classification, analysis, and retrieval of cause-of-death information.Footnote 22 These were major initiatives, requiring major investments, and they built an important foundation for modernising mortality statistics infrastructure more broadly. But the newly computerised nosology systems did nothing to address the larger issues of incomplete or inaccurate data at the source.
As automation crusades intensified in the 1970s and 1980s, the discourse around vital statistics modernisation reflected the era’s optimism about technology. With vital records infrastructure using technologies like microfilm and magnetic tape files, state health officials could send data to their federal counterparts quickly, freeing workers from the former slog of having to produce photocopies or, even more tedious, duplicate handwritten copies.Footnote 23 Some officials also saw modernised infrastructure as a strategy to counter the rise of budget cuts.Footnote 24 At one 1989 public health conference, for example, an official from Rhode Island explained that the state’s newly enhanced capability to obtain quality public health statistics was a direct result of the microcomputer revolution; the department no longer had to rely on mainframes, which made modernisation more feasible financially. ‘Were it not for the microcomputer and the people who knew how to use them, our statistical systems would have fallen into genuine disarray when the funding cuts came’, the official said.Footnote 25 In other words, while constraints surged, public health statisticians breathed a sigh of relief, trusting that advancements in technology were coming just in the nick of time.
The challenges of a paper-based process
Despite real progress, however, by the early 1990s, vital statistics officials agreed that recent advancements to U.S. public health data systems remained insufficient. One major report on vital statistics stated that the country’s data infrastructure ‘clearly has not kept pace with the growth in our health system’s size and complexity’, adding that data systems were becoming ‘increasingly inadequate to perform the necessary evaluations’.Footnote 26 In another report, a state-level director of health statistics identified labour as a key impediment to technological progress: ‘It is clear that there is a need for automation in registration programs as record volumes grow, demands for services increase, and resources and funds decline’, the official wrote, but the arduous work of gathering the data still posed a major challenge. ‘This obviously has to be a main concern and one that is resolvable prior to committing resources to an automation plan’, the official added.Footnote 27
These data pain points were especially acute in the fragmented, laborious system of registering deaths. What makes timely death registration so difficult is that the information comes from multiple sources: funeral directors work with the deceased’s family and friends to gather demographic data, while cause-of-death data comes from a doctor at a hospital, or, in other cases, from a coroner or medical examiner. Rather than this information being collected on paper by the funeral director, then shipped off to local health officials, who then send the data to state and eventually federal officials, increasingly, public health officials agreed it was time for records instead to be ‘created, edited, coded, queried, and corrected at the source in an electronic format’.Footnote 28 This idea that death registration could begin at the data’s source – in the offices and hallways of funeral homes and hospitals – became a methodological imperative of modernisation efforts.
Another reason why U.S. vital statistics officials began pushing for electronic death registration was simple: they had a successful model in electronic birth registration. The decade prior, in 1980, researchers at the University of California, Santa Barbara, collaborated with county health officials to develop a system that could produce structured birth data, vet the data automatically, then send it to local and state vital statistics officials electronically.Footnote 29 This new system not only optimised the speed at which officials learned about births but also reduced clerical staff demands and integrated quality checks directly into the process of collecting the data, rather than relying on a separate vetting step after collection. Given these affordances, and its relatively quick implementation trajectory, public health officials considered electronic birth registration a massive success; by 1995, nearly 70% of all U.S. births were registered electronically.Footnote 30
But officials also knew that electronic death registration would be ‘much harder to achieve’ than the parallel process for births, as it hadn’t been technology alone that led to the success of electronic birth registration.Footnote 31 As historian Susan J. Pearson explains, for U.S. birth registration to produce meaningful results, the system required ‘not just laws, forms, and delegated duties but also ideological investment in registration as a public good’.Footnote 32 This ideological investment had not been easy to come by: child welfare reformers and other public health boosters had spent decades organising and funding initiatives to champion birth registration, a groundwork laid long before the system’s computerisation. Moreover, as officials well knew, the data collection needs of birth registration tended to be more straightforward than those of death registration, with data coming from only one source, rather than from multiple sources. Put simply, though there were certainly similarities between the two systems, the process of digitising death registration would be far more involved than the process had been for births.
As vital statistics officials turned towards electronic death registration, the question of data collection labour became a real threat to modernisation. For the new system to work, funeral directors, physicians, coroners, medical examiners, and local registrars needed to grasp that their interactions with EDRS technologies determined the legitimacy of population-level mortality statistics. This would be challenging, officials knew, because many death-care workers prioritise the living, not the dead.Footnote 33 They are the doctors bustling through the hallways of hospitals, working frantically to keep their still-living patients alive; they are the funeral directors who spend long hours with grieving families, providing ritual, solace, and closure. Though committed to the technology’s inevitability, vital statistics officials nonetheless knew the new system’s pivotal test would be social, rather than technical. Officials would need to convince physicians and funeral directors to understand the importance of death registration data, to see themselves also as data workers.
Technology as the solution
To address the problem of inadequate death certificate data, U.S. vital statistics officials focused initially on what seemed like a straightforward solution: more physician training. Not unlike the ideological investment campaigns that accompanied the rise of birth registration, officials agreed that what death registration needed now was more awareness and education. In 1989, the NCHS and National Committee on Vital and Health Statistics (NCVHS) held a workshop on improving cause-of-death statistics with the goal of building more collaborative channels between government officials and the various professional organisations representing physicians, hospitals, and medical schools. The workshop produced a set of recommendations that called for widespread educational efforts to train both medical students and practicing physicians on ‘the importance of cause-of-death statistics and their value in improving the practice of medicine, as well as the mechanics of completing the death certificate’.Footnote 34 Indeed, this idea of more education continued throughout the 1990s. As one forensic pathologist wrote, the success of any future death registration system ‘will depend on a greater commitment from certifiers of death than has been displayed by them in the past’.Footnote 35
But the road ahead was steep. Even beyond the realm of death paperwork specifically, the challenges of data entry labour only intensified as patient records more broadly became computerised.Footnote 36 Many physicians in the early 1990s saw data entry as ‘unworthy of their time or consideration’, with the inadequacies of early hospital computer systems turning data entry tasks into ‘exercises of advanced frustration’, as some physicians put it.Footnote 37 By the 1990s, two major reports highlighted the extent of the issue. The first, a report from the Institute of Medicine outlining a new vision for computer-based patient records, stated: ‘Perhaps the single greatest challenge that has consistently confronted every clinical system developer is to engage clinicians in direct data entry’.Footnote 38 The report noted that what the system needed was not additional technological breakthroughs but rather breakthroughs of ‘human interface – the place where man and machine meet’.Footnote 39 In the second report, published by the Government Accountability Office on the rise of automated medical records, the agency found that physician resistance was considered by many to be the ‘major obstacle’ to technological progress.Footnote 40 Resistance stemmed in part from general hesitations about technology, but the report also noted that some physicians were dubious of potential effects on the underlying data, in that rigid software might impose predefined constraints on patient information.Footnote 41
It did not take long for public health officials to recognise that if developed and implemented well, EDRS could solve not just the problem of sluggish death data transfer but also the problem of poor data quality. At another workshop on improving cause-of-death statistics, held in 1991, officials called for ‘a total reevaluation of how medical information is collected through the death certificate’.Footnote 42 The group suggested building interactive and user-friendly software for death information collection, much like the electronic birth registration systems were doing. The software should be uniform in nature, attendees decided, able to be used across jurisdictions. It should have individual modules for its various users, like doctors, coroners, medical examiners, funeral directors, and local registrars. Most importantly, the software should ‘contain feedback features, including prompts’, so that data quality could be vetted immediately and at the source, rather than after some delay. Finally, the group decided that design requirements should be sketched out within two years, with the medical module operational by 1996. The workshop concluded with top officials offering a hearty round of congratulations to attendees for their ‘very courageous’ efforts and ‘very constructive and innovative ideas’.Footnote 43
With momentum in full force and a plan now largely in place, initiatives to reengineer U.S. death data collection spun up throughout U.S. public health agencies and organisations. In 1994, the NCHS joined forces with the National Association for Public Health Statistics and Information Systems (NAPHSIS) to create the Steering Committee to Reengineer the Death Registration Process, a group tasked with examining the feasibility of overhauling death registration. After meeting six times between 1994 and 1996, the committee produced a final report that embraced the future of U.S. electronic death registration fully. Advancements in computer technology had ‘made possible more efficient methods of capturing, processing, and communicating death data’, the report said, and the country’s commitment to achieving faster and better death certificate information had been renewed.Footnote 44 The report contained a broad set of system design, legislative, and implementation recommendations. More granularly, too, to provide clear and streamlined guidance, the report included a list of eight minimum technical attributes that any EDRS should include.
By the end of the 1990s, U.S. vital statistics officials across federal, state, and local government had latched onto the moment’s optimism, committing to equip the country’s death registration jurisdictions with sophisticated mortality statistics systems for the new century. As media studies scholar Jacqueline Wernimont describes, ‘fierce headwinds of techno-solutionism’ permeated many of these discussions, with officials aware of the data entry labour problem yet nonetheless convinced by the software’s ability to resolve it.Footnote 45 Consciously or not, public health officials likely could not see an alternative to techno-solutionism; as the twenty-first century dawned, officials grew increasingly alarmed by the lasting inadequacies and inefficiencies of paper-based death registration, understanding full well that the country’s public health surveillance systems needed dramatic improvements. ‘Up to now, the registration activities have been a low priority’, one official wrote in 1997. ‘As long as a copy of a certificate could be prepared, the system was thought to be adequate. This is no longer the case’, he added, pointing to the increased documentation requirements needed for public service support, the increased mobility of the U.S. population, and ‘decreases in program staff due to serious budget deficiencies’.Footnote 46
Indeed, the techno-solutionism that fuelled early EDRS efforts only intensified as the NCHS’s funding situation became more and more dire, the financial plight growing in tandem with the motivation for widespread automation.Footnote 47 Precarious funding situations constrained the NCHS throughout the 1990s, compromising vital statistics collection efforts; budget cuts even forced the agency mid-decade to drop the collection of some metrics on births and deaths temporarily and to discontinue collecting individual record data on marriage and divorce.Footnote 48 In this context, the 1990s-era commitment to new technology in U.S. vital statistics became a kind of last-ditch attempt. ‘Our resources are continuing to decline, so we thought an electronic system was certainly the way to proceed’, said a state-level health official at the end of the decade.Footnote 49 With financial pressures mounting, vital statistics officials were firm in their belief that electronic death registration was both necessary and necessarily imminent, going so far as to predict that the new systems would be in place in most U.S. states by the year 2000.Footnote 50
The problems with pick lists
In 1997, when the Steering Committee to Reengineer the Death Registration Process produced its report of recommendations for EDRS developers, the group kept legislative and implementation guidance high-level but offered a slate of development and design recommendations. The eight minimum technical attributes as envisioned by the committee were as follows: (1) all systems must be able to retain original text, (2) systems should have built-in tutorials and help-screens, (3) systems needed to be designed in a way that rendered the data compatible with vital statistics coding specifications, (4) systems needed sufficient edit and query functions, (5) systems should be able to transmit uncoded data to the NCHS for further coding, (6) systems needed to meet reasonable standards of security, along with (7) reasonable standards of user authentication, and, finally, (8) systems needed to be able to generate printed death certificates that conformed to the jurisdiction’s specific legal standards.
To be sure, each minimum technical attribute posed its own challenges, but it was the broader idea that EDRS tools needed both specificity and flexibility that became almost impossibly complicated. According to the committee’s descriptions, each system at a minimum had to be compatible with the NCHS’s instruction manuals, though systems were also supposed to be ‘sufficiently flexible to meet the needs of the states’, as the states were legal jurisdictions that could impose their own additional data demands, asking for information on top of the metrics required by federal officials.Footnote 51 ‘When appropriate’, the report said, ‘coded information should be generated from selection lists presented to the user during the data entry process’.Footnote 52 In other words, the committee members recommended that EDRS developers use built-in help features, which could guide funeral directors and busy physicians through data entry in a way that was sure to meet the needs of all officials, federal and otherwise.
This idea of selection lists – pre-determined sets of available data options, also called pick-lists, pick lists, or dropdown lists – became a precarious issue, an attempt to navigate the tension of creating a system that could enhance the quality of death data while also easing the experience of data entry. This tension appeared conspicuously throughout the Steering Committee’s report. In general guidance about how jurisdictions should design the technologies, for example, the report said that an EDRS ‘should provide maximum accessibility for data entry’; a later recommendation then said systems should ‘maximise data quality’.Footnote 53 Additionally, the report recommended that data providers be alerted automatically by the EDRS upon the input of inconsistent information, and that there should be built-in help features if users found themselves unsure of how best to enter information. ‘When appropriate, selection lists should be used to provide assistance in the completion of items. However, the lists should be used in a manner to ensure that the actual literal response is retained in the electronic record’, the report said.
Given these tensions, the Steering Committee’s report launched yet another round of discussions about how best to design and roll out the country’s network of EDRS technologies. In 1999, the NCHS convened dozens of experts on death data, a group known as the International Collaborative Effort (ICE) on Automating Mortality Statistics. The ICE group had met once before, in 1996, to discuss other death data initiatives, but it wasn’t until this second gathering that the topic of EDRS took off, with officials referring to electronic death registration as ‘just on the brink’ of implementation.Footnote 54 As attendees explained, EDRS held significant promises, with the tools able to reduce data entry demands while also improving the ‘very burdensome’ paper-based workflow and the quality of the data itself.Footnote 55 Yet the topic of data entry still loomed over discussions. Echoing earlier concerns, attendees spoke of the disconnect between the actual value of death data and the perceived value with which doctors and funeral directors regarded it. As one medical epidemiologist bemoaned, vital statistics officials had made little progress on getting EDRS users to see themselves as data workers; physicians specifically, the epidemiologist said, still ‘do not realize that the data relies on them’.Footnote 56
At the next gathering of the ICE group, a follow-up event held in 2003, organisers dedicated an entire session to electronic death registration, referring to the topic as ‘a very high priority’.Footnote 57 Once again, issues raised about EDRS concentrated on the technology’s users. One official noted that though some viewed EDRS as ‘a virtual panacea for the persistent problem of training physicians on how to complete death certificates’, the benefits would be realised only if it was the physician completing the certification.Footnote 58 There was the risk that medical records staff at hospitals, like medical scribes or data clerks, might be tasked with data entry, rather than it being handled by the physicians who were trained – and in some cases, obligated legally – to perform the work. ‘I witnessed this possibility in one of the prototype electronic death registration systems that I visited’, the official cautioned, noting that the computers set up for cause-of-death reporting had been placed ‘in the medical records section of the hospital’, rather than in easy proximity to the doctors.Footnote 59 Certainly, with the paper-based process, there had been risk that non-physicians provided the data, but electronic death registration raised the stakes of the consequences, the official explained, as untrained staff inputting data into an EDRS ‘would negate the potential benefits’ of modernising the technology.Footnote 60
More pressingly, the 2003 ICE convening elevated warnings about the still unresolved challenge of optimising for user friendliness without sacrificing data quality. EDRS might inadvertently make the death registration process seem ‘too easy, indeed misleadingly easy’, one official warned.Footnote 61 Of particular concern was the topic of pick lists – those pre-determined sets of data options from which EDRS users would be forced to choose. Though pick lists had been recommended by the earlier Steering Committee for use ‘when appropriate’, the committee had not specified what constituted appropriate use. To be sure, pick lists might speed up the process of registering a death, encouraging doctors to use EDRS, but the lists left little opportunity for the input of data that did not fit easily into existing buckets. ‘Pick lists are widely used in electronic questionnaires and forms, so it is not surprising that energetic software developers have incorporated such lists into hospital information systems’, an official noted, adding that software developers ‘could not be expected to be aware of the potentially damaging impact of such lists on mortality data precision and accuracy’.Footnote 62 Practically, too, pick lists would impose regular maintenance demands on the software, to keep up with shifting medical terminology. And decisions about terminology changes would, in turn, require more human labour.
Conclusion
On top of the behemoth challenge that was optimising for both data quality and user friendliness, other obstacles arose in the early 2000s, too. Funding, for example, became a major issue. Y2K preparations had run many health department technology budgets dry, and in the aftermath of 9/11, cuts further strained many states.Footnote 63 In response, jurisdictions began to lean on external partners, like the Social Security Administration (SSA), to fund EDRS developments.Footnote 64 This financial support was critical; one news article went so far as to summarise the partnership by implying the SSA’s investments kept the vision of electronic death registration ‘still alive’.Footnote 65 But the SSA’s involvement in EDRS development also broadened the system’s stakeholder base, extending the goals of EDRS and shifting the incentives of data collection.Footnote 66 Another issue was in finding the technical proficiency needed to oversee the software development at scale. As a result of insufficient technical expertise, many jurisdictions pushed forward initial projects that merely automated existing paper-based processes, rather than reengineering the system fully.Footnote 67
Taken together, these challenges plagued EDRS efforts, leading to uninspired progress and even some material consequences on the data itself. As a result of the lagging EDRS rollouts, for example, many jurisdictions were forced to delay implementation of the 2003 death certificate revisions, hampering their ability to further standardise the data and improve its quality.Footnote 68 Moreover, as the years went on, NCHS officials continued to grapple with questions about the most effective way to retrieve cause-of-death information, how best to connect funeral directors and physicians securely online, whether and how to take advantage of the software systems already used by funeral directors and medical professionals, and the level of specificity EDRS prompts should pursue, so as to not render the tools ‘counterproductive’.Footnote 69 Despite major improvements to mortality statistics systems in the first half of the twentieth century, and despite the decades of investments that followed, the United States entered the new century with information moving slowly and clumsily through the country’s fragmented death data infrastructure. Indeed, the growing pains of electronic implementations challenged U.S. death registration temporarily but profoundly.
The persistence of EDRS’s key tension – optimising for both data quality and ease of data entry – was due neither to a lack of awareness nor to a lack of effort; throughout the history of discussions about the need to digitise death registration, vital statistics officials were aware of the issues that undermined the technology’s potential for success. Yet they were nonetheless unable to push forward EDRS tools that addressed both challenges fully. As one New York City official wrote after the city tried – and failed – initially to roll out its EDRS, the system’s users are the ‘physicians, nurse midwives, medical examiners, coroners, funeral directors, hospital and clinic directors, clerks, and temporary staff who complete the worksheets and certificates that eventually become data’.Footnote 70 Continuing the decades-old efforts to expand and improve death registration technology remained critical, he added, but at the end of the day, the electronic systems ‘still rely on careful and accurate collection and entry of data’.Footnote 71
Competing interests
The author declares none.
Open access
